Sie sind auf Seite 1von 11

SPE 168606

A New Method for Fracture Diagnostics Using Low Frequency


Electromagnetic Induction
Saptaswa Basu, SPE, and Mukul M. Sharma, SPE, The University of Texas at Austin

Copyright 2014, Society of Petroleum Engineers


This paper was prepared for presentation at the SPE Hydraulic Fracturing Technology Conference held in The Woodlands, Texas, USA, 46 February 2014.
This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents of the paper have not been
reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect any position of the Society of Petroleum Engineers, its
officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written consent of the Society of Petroleum Engineers is prohibited. Permission to
reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.

Abstract
Currently, microseismic monitoring is widely used for fracture diagnosis. Since the method monitors the propagation of shear
failure events, it is an indirect measure of the propped fracture geometry. Our primary interest is in estimating the orientation
and length of the propped fractures (not the created fractures), since this is the primary driver for well productivity. This
paper presents a new Low Frequency Electromagnetic Induction (LFEI) method which has the potential to estimate not only
the propped length, height and orientation of hydraulic fractures but also the vertical distribution of proppant within the
fracture.
The proposed technique involves pumping electrically conductive proppant (which is currently available) into the fracture
and then using a specially built logging tool that measures the electromagnetic response of the formation. Results are
presented for a proposed logging tool that consists of three sets of tri-directional transmitters and receivers at 6, 30 and 60
feet spacing respectively. The solution of Maxwells equations shows that it is possible to use the tool to determine both the
orientation and the length of the fracture by detecting the location of these particles in the formation after hydraulic
fracturing. Results for extensive sensitivity analysis are presented to show the effect of different propped lengths, height and
orientation of planar fractures in a shale environment. Multiple numerical simulations, using a leading-edge (FEKO)
electromagnetic simulator, indicate that we can detect and map fractures up to 250 feet in length, 0.2 inches wide, and with a
45 degree of inclination with respect to the wellbore. Special cases, such as proppant banking and wells with steel casing in
place, were also considered.
Introduction
Hydraulic fracturing has become a major driver for oil and gas production in United States land operations. Knowledge of
hydraulic fracture dimensions is of great importance when it comes to predicting production and validating reservoir and
fracture models; and utlimately, improving production economics. However, we still lack a cheap, direct and repeatable postfracturing diagnostic tool to measure the dimensions of hydraulic fractures. Cipolla and Wright (2000) evaluated current
fracture diagnostic tools and summarized the capabilities and limitations of each method, as shown in Table 1.
Microseismic monitoring, which is the most commonly used technique in the industry today, fails to give us the effective
propped length or the proppant distribution within hydraulic fractures. These two parameters have a first order impact on
production. Microseismic inversion is also dependent on the velocity model used. The velocity model can change during the
fracturing job as stress reorientation takes place with each fracture stage, as demonstrated by Roussel and Sharma (2011).
Therefore, there is room for a new technology in this domain that addresses these issues and gives us a more robust answer.
Symington et al (2006) proposed the Electrofrac method of heating formations in-situ, which uses conductancts (calcined
coke) for proppants. These conductants have an electrical resistivity high enough for resisitive heating but low enough for
sufficient electric current conduction. This process is used for converting oil shale to producible oil and gas. Another
application, which has not been explored by the industry yet, could be of determining the location of these conductive
proppant using the difference between the conductivity of the rock matrix and the proppants. In this paper, we propose using
an electrically conductive proppant (conductant) for hydraulic fracturing and then using a LFEI tool for determining the
propped length of fractures. We will first study the feasibility of the proposed technology by solving a few simple forward

SPE 168606

problems. Then we will proceed to do some sensitivity analyses to determine the best frequency of operation, transmitterreceiver spacing and the limitations of this tool in different rock matrix and proppant types/properties.

Table 1 - Capabilities and Limitations of Current Fracture Diagnostic Tools (after Cipolla and Wright, 2000)

There is a significant body of work on conventional electromagnetic induction tools in different geologies. Anderson
(2001) affirms that anisotropy, from the point of view of a logging tool, depends on scale. For example, in addition to
microscopic anisotropy occurring at particle scale, a formation consisting of a series of isotropic beds of different lithology
also behaves anisotropically if the logging tool is significantly longer than the bed thickness. Moran and Gianzero, (1979),
proposed a technique for measuring anisotropy using a combination of horizontal and vertical coils since vertical coils are
sensitive to Rv (effective vertical resistivity). Alumbaugh et al (2001) also suggested that in regions of complex geology it is
necessary to use both a horizontal and a vertical magnetic field sensor and employ multiple source-receiver offsets.
Therefore, we have modeled our simulations using tri-directional magnetic dipoles, to get the complete response reflecting
the anisotropic nature of electrical conductivity hydraulic fractures.
The method of Moran (1979) is currently the most widely used in conventional induction logging. It considers a
homogeneous formation of conductivity with a loop of formation conductivity + . This solution is analogous to the
Born approximation in quantum mechanics and involves a single scatterer in the same manner, therefore, it is often called the
Born response function. However, experience has shown that it is more informative to study the integrated radial response

SPE 168606

function, which gives the contribution from a cylindrical volume of formation (such as the invaded zone) rather than just a
thin shell. We have also followed this methodology while designing our simulation model and meshing algorithm.
Anderson (2001) investigated the inversion of conventional triaxial electromagnetic tools (frequency of operation: 20
kHz) in layered media and showed that the average radius of investigation corresponds to approximately half the transmitterreceiver coil spacing. Also, this tool is sensitive to vertical and horizontal resitivities and dip. Alumbaugh and Wilt (2001)
demonstrated the numerical feasibility of generating a 3D image of the region surrounding the borehole using a vertical
magnetic dipole (VMD) source. Tang et al (2006) have shown that the longest fracture a conventional array induction tool
(AIT) can detect is 40 ft.
Low-frequecny electromagnetic induction for anomaly detection has been used outside the energy industry also.
Milesevic et al (2011) used LFEI (50 Hz) to calculate induced voltages in an underground pipeline (240 m) using EMTPATP software. Dorn (2002) used low frequency electromagnetic induction (1 kHz) for shape reconstruction at a depth of
investigation of 200 m. Their method involved using level sets and adjoint fields to solve Maxwells equation in the
frequency domain. In this paper we intend to marry the above approaches and design a LFEI triaxial logging tool that can
detect hydraulic fractures up to 250 ft propped half-length (median depth of investigation) along with its orientation in the
formation.

Method
In this paper, we solve the forward problem (E field signal received) for a given fracture propped length and orientation in
horizontal and vertical wells. We model the hydraulic fractures as conductive ellipsoidal disks in an explicit non-conductive
borehole (except for cased holes), as shown in Fig. 1 for ease of studying first order effects. After solving the simplest cases,
we move on to more complex cases of wellbore with steel casing, and proppant banking in horizontal wells. We use FEKO,
an electromagnetic simulator that solves Maxwells equation (Eqn. 1 and Eqn.2) in the frequency domain. In the frequency
domain, Maxwells equation can be solved by the Method of Moments (MOM), Physical Optics (PO), Geometric Optics
(GO), Geometric Theory of Diffraction (GTD) and Uniform Theory of Diffraction (UTD). The advantage of using the
Method of Moments (MOM) is that they are free of dispersion errors, no artificial boundary is needed for exterior problems
and only volumes needs to be discretized in 3-D space. The other methods are based on asymptotic high-frequency
expansions of the equation and are, therefore, best suited for high frequency problems.
E= MjH
H = J + jE

(1)
(2)

We model the rock matrix and hydraulic fractures as volumes with different electrical resistivities. We have used the
Method of Moments (MOM) with a Volume Equivalence Principle (VEP) as the the solution method. VEP is one
implementation of MOM that is more accurate at low frequencies. It discretizes the volumes into tetrahedra, which is
computationally more expensive, but gives a more accurate solution. We have also incorporated low-frequency stabilization
and double precision in our numerical solution. Mesh refinement is used to create a finer mesh near the tri-directional
sources, where the field gradient is higher. Since MOM forms a dense matrix as part of its solution process, the traditional
implemention of MOM scales poorly both in memory and runtime requirements. Therefore, our simulation implements an
Adaptive Cross-Approximation (ACA) method.
In this paper we focus our attention on a LFEI tool with one triaxial transmitter and three triaxial receiver sets spaced at 6
ft (1.82 m), 30 ft (9.144m) and 60 ft (18.288m) respectively. The detailed justification and reasons for selecting this
transmitter and receiver spacing are provided in Salies (2012) and Basu (2014). We model induction coils as vertical
magnetic dipoles (VMD) of 1000 A-m operating over a range of frequencies from 1 to 106 Hz. Salies (2012) plotted depth of
penentration against frequency of operation for a lossy medium as shown in Fig. 2. A bucking coil to cancel out direct
transmitter to receiver coupling (primary field) has also been incorporated. So we only observe the secondary field which
contains information about the conductivity and geometry of anomalies in the formation matrix.

SPE 168606

Fig. 1 Geometry of fractures simulated in the model.

Fig. 2 Frequency vs depth of penentration (after Salies,


2012)

We simulate 6 different cases which demonstrate different facets and applications of this proposed technology. Table 2
lists the parameters used in the Base Case simulations.
Parameter
Matrix

Value
5 Ohm-m

Proppant

0.001 Ohm-m

Steel casing

7 10-7 Ohm-m
1000 A-m

Strength of VMD
Frequency of operation

100 Hz

Table 2 - Parameters for simulation of the Forward Problem

Case 1: Vertical Fracture in Vertical Well. In this case we model a vertical fracture in a vertical well with the fracture
propagating along the Y axis as shown in Fig. 3. Keeping all other factors same; fracture half-lengths of 15m, 30m, 50m and
75m were studied. The Z-signal in the long spacing receiver was most sensitive to change in fracture length, and increases
proportionally with increasing fracture length as shown in Fig. 4. The medium and short spacing signal gets contaminated at
higher half-lengths, but gives us the location of the fracture at the wellbore effectively, by responding to near wellbore
conductivity changes.

Fig. 3 Model vertical fracture

Fig. 4 Reciever response to vertical fracture in vertical well

Case 2: Oriented Fracture in Vertical Well. Keeping the fracture half-length constant (30 m) we rotate the fracture around
the axis of the borehole so that it makes some angle with the Y axix. Based on our simulation results it is evident that we
can find the angle from the ratio of the signal in X and Y direction as shown in Fig. 5. The half-length can be independently
tracked using the received signal in the Z direction as in Case 1. We can resolve fracture angles of upto 45 degrees uniquely
using this technique. For larger angles the pattern repeats itself and the the ratio of the E fields becomes multi-valued.

SPE 168606

10.00

Ey/ Ex

8.00
6.00
4.00
2.00
0.00
0

20

40

60

80

100

Angle in degrees
Fig. 5 Reciever response to oriented fracture in vertical well

Case 3: Orthogonal Fracture in a Horizontal Well. In this case we model penny- shaped fractures which are perpendicular
to the axis of the borehole. We start off with a fracture of half length 30m as shown in Fig 6, with the hydraulic fracture
located at 30m from the starting point of the tool. We note the signal received in the Z directional in the 6ft (short spacing)
receiver, 30ft (medium spacing) receiver and 60ft (long spacing) receiver and plot for each depth as the tool moves along the
borehole. The response to the conductive proppant filled hydraulic fracture is obvious in this simulated log. We repeat the
same process for fractures of half-length 50 m, 75m and 100m and plot similar logs as shown in Fig 7. There are two key
observations: (a) there is a monotic increase of receiver signal with increasing fracture half length and (b) the short and
medium spacing receiver signals get noisier with increasing fracture length, while long spacing signal remain more
consistent.

Fig. 6 Schematic showing a low frequency logging tool, a hydraulic fracture and the wellbore.

SPE 168606

(a)

(b)

(c)

(d)

Fig. 7 - Receiver response to fractures placed orthogonal to a horizontal well for different fracture lengths (a) 30m, (b) 50m, (c) 75m
and (d)100m.

Case 4: Non-Orthogonal fracture in a Horizontal Well. We implement the same method as in the previous case, but now
we tilt the fracture in angles of 15 degree, 30 degree and 45 degree in an axis perpendicular to the axis of the borehole as
shown in Fig. 8(a). We map the signal received in the tridirectional long spacing recivers in a 3D plot to diagnose the effect
of a particular angle for fracture half-lengths of 30m, 50m and 75m as shown in Fig. 8(b). We observe that the received
signal increases monotonically with increasing fracture half length for each angle. Also, each angle gives a non-intersecting
curve in the 3-D plot which means if we have prior knowledge of the angle, we can find the half-length and vice versa. If
both are unknown, it gives us a non-unique solution that demands an inversion algorithm that is being investigated.

(a)

(b)

Fig. 8- (a) Schematic showing wellbore with fractures at angles of 0, 15, 30 and 45 to normal, respectively (left to right) and (b)
Receiver response to oriented fractures in a horizontal well

SPE 168606

Case 5: Proppant distribution in a Penny Shaped Fracture in a Horizontal Well. In this case we consider fractures which
are partially filled with water (blue) while the rest of the fracture is occupied by proppant (green). Fig. 9 shows the five
different configurations of proppant bank considered and Fig. 10 shows the distinct signature of each configuration. This set
of simulations shows that we can distinguish between different propannt distributions using LFEI as it is sensitive to the
electrical properties of the proppant. However, proppant distribution coupled with complex geometries of the fracture can
give rise to non-unique solutions of the inverse problem. This is a common issue in many inverse problems in well logging
and needs complex inversion algorithms to address. This aspect of the problem is currently being investigated.

Fig. 9 Schematic of different proppant distributions in the fracture that were modeled.

Fig. 10 Tridirectional receiver response to 5 different proppant distributions.

Case 6: Steel Casing.


If we employ steel casing in each of the above configuration and re-run the simulations, we observe that in each of the cases
the signal quality improves. Fig. 11 shows the case of an orthogonal fracture in a horizontal wellbore with steel casing
installed. Pardo et al (2013) have also reported the same independently using their own code, and have theorized that it can be
due to the casing acting as a long electrode.

SPE 168606

Fig. 11 Receiver response to orthogonal fracture in a horizontal well with casing installed.

From the above simulation cases we can see that LFEI is capable of measuring fracture propped half length, height, width,
dip, azimuth as well as detect proppant distribution in both cased and un-cased wellbores. It is independent of any fracture/
reservoir/ geomechanical model and directly measures the location of the proppant in the fracture.
Sensitivity Analysis
We have run the following sensitivity cases to determine the best conditions to use LFEI for fracture diagnostics. For each
case, we have kept all other factors constant, while changing the one under investigation. For the pupose of this sensitivity
analysis, we have kept the fracture half length constant at 30m in Case 1 scenario, which changing each of the following
parameters.
Frequency of Operation
We varied the frequency of operation from 10 Hz to 1 MHz in logarithmic increments of 10. We observe (Fig. 12) that the
best signal is received for a frequency of 100 Hz.

Fig. 12 Sensitivity analysis for Frequency of Operation

SPE 168606

Transmitter-Receiver Spacing.
We simulated 8 different cases of transmitter-receiver spacing as shown in Fig. 13. We observe that (a) the 60 ft spacing has
the highest response at the fracture but is also fairly noisy, and, (b) shorter spacing is most sensitive to the location of the
fracture in the borehole. Therefore, we concluded that we need both the short and long spacing to find the location and the
dimensions of the fracture. We can combine the readings for all three receiver arrays and find the exact location of the
fracture and dimensions using Phasor processing (Anderson, 2001).

Fig. 13 Sensitivity analysis for transmitter-reciever spacing

Resistivity of the Rock Matrix.


Keeping all other parameters the same, we increased the resistivity of the matrix from 1 ohm-m to 500 ohm-m. From Fig. 14
we observe that increasing the matrix resisitivity increases the received signal, which implies that having a higher matrix
resistivity would also increase the depth of investigation. Also we can infer that the tool will not lose its functionality within
this range of matrix resistivity.

Fig. 14 Sensitivity analysis for rock matrix resistivity.

10

SPE 168606

Resistivity of the Proppant.


Although we have several electrically conductive proppants available in the market, in this section we explore the possibility
of finding the best proppant for our application. As demonstrated in Fig. 15, we change the resistivity of the proppant and
observe that the lower the resistivity of the proppant (higher the contrast of resistivity with matrix), the higher the received Efield signal. However, if we decrease the resistivity beyond 0.0001 ohm-m, the received signal is insufficient and cannot be
recoded reliably. Also increasing the resistivity beyond 1 ohm-m makes the overall signal so noisy that it is difficult to detect
the hydraulic fracture. Ideally the rock resistivity should be low enough to provide sufficient current in the rock matrix and
high enough to get a sufficient reistivity contrast with the conductive proppant. Fortunately these criteria are satisfied over
the entire range of resistivities for typical rocks.

Fig. 15 Sensitivity analysis for conductivity of proppant

Conclusion
In this work, we propose that a low frequency induction logging tool can be used to detect hydraulic fractures propped with
conductive proppants. We have used realistic parameters to simulate the geology, the proppant and the formation (Table 2).
We can summarize our findings as follows:

A new method is proposed that can estimate the length, orientation and height of propped hydraulic fractures with
propped half-lengths upto 75m (250 ft) in both horizontal and vertical wells.

In addition to the fracture dimensions and orientation, the method has the potential to provide an estimate of the
vertical distribution of the proppant.

We show how the use of three receivers with tri-directional antennas can be used to infer fracture geometry in many
commonly encountered fracture geometries.

A transmitter-receiver spacing of 60m helps us get a deep reading. However, we also need shorter and medium
spacing receivers for better mapping of the fractures.

Based on our simulations, 100 Hz appears to be the frequency most suitable for this application.

The proposed tool is shown to work well for a range of matrix resisitivties between 1 ohm-m and 500 ohm-m.

The presence of casing improves the signal to noise ratio received at the receivers and allows even better fracture
diagnostics than for the no-casing cases.

Acknowledgement
The authors would like to acknowledge the support provided by the companies sponsoring the JIP on Hydraulic Fracturing
and Sand Control at the University of Texas at Austin (Air Liquide, Air Products, Anadarko, Apache, Baker Hughes, BHP
Billiton, BP, Chevron, ConocoPhillips, ExxonMobil, Ferus, Halliburton, Hess Corporation, Linde Group, PEMEX, Pioneer
Natural Resources, Praxair Inc., Saudi Aramco, Schlumberger, Shell, Southerwestern Energy, Statoil, Weatherford, and
YPF).

SPE 168606

11

References
Alumbaugh, D.L. and Wilt, M. J., 2001. A numerical feasibility study of three-component induction logging for three-dimensional imaging
about a single borehole. Petrophysics 42 (1):19-31.
Anderson, B.I., Barber, T. D., and Gianzero, S. C., 2001. The effect of crossbedding anisotropy on induction tool response. Petrophysics
42 (2): 137-149.
Bertete-Aguirre, H., Dorn, O., Berryman, J.G., et al. 2002. 3D-electromagnetic imaging using adjoint fields, IEEE Antennas and
Propagation Society International Symposium., Vol.2:773-776. doi: 10.1109/APS.2002.1016761.
Cipolla, C.L. and Wright, C.A. 2000. Diagnostic Techniques to Understand Hydraulic Fracturing: What? Why? And How? Paper SPE
59735 presented at the SPE/CERI Gas Technology Symposium, Calgary, 2-5 April.
Milesevic, B., Filipovic-Grcic, B.; Radosevic, T., 2011. Analysis of low frequency electromagnetic fields and calculation of induced
voltages to an underground pipeline. Proceedings of the 3rd International Youth Conference on Energetics (IYCE). Vol 1: 1-7. Leiria,
Portugal, 7-9 July.
Moran, J.H. and Gianzero, S.C., 1979. Effect of formation anisotropy on resistivity-logging measurements: Geophysics, Vol 44: 12661286.
Pardo, D. and Torres-Verdn, C., 2013. Sensitivity analysis for the appraisal of hydrofractures in horizontal wells with borehole resistivity
measurements. Geophysics 78(4), D209D222.
Roussel, N.P. and Sharma, M.M. 2011. Optimizing Fracture Spacing and Sequencing in Horizontal-Well Fracturing. SPE Prod & Oper 26
(2): 173-184. SPE-127986-PA. doi: 10.2118/127986-PA.
Salis, Natlia Gasto, 2012. Study on the feasibility of using electromagnetic methods for fracture diagnostics. MS Thesis, University of
Texas, Austin (August 2012).
Symington, W.A., Kaminsky, R.D, Meurer, W.P., et al. 2010. ExxonMobils Electrofrac Process for In Situ Oil Shale Conversion. In Oil
Shale: A Solution to the Liquid Fuel Dilemma. Chap. 10, 185-216. ACS Symposium Series. doi:10.1021/bk-2010-1032.ch010
Tang, Y., T. Wang, B. Corley, and J. Morrison, 2006, Interpreting fracture length from multi-array induction logs. Proceedings of 76th
Annual International Meeting, SEG, Expanded Abstracts 25, 421425.

Das könnte Ihnen auch gefallen